This review shows that studies claiming vision is not a factor in dyslexia may suffer from poor specificity or sensitivity

1. Dyslexia & Vision: A Critical Review
By Stuart Warren B.Optom, MSc(Optom)

2. Introduction
It has been known for decades that dyslexics make more erratic eye movements when
reading compared to non-dyslexics. In addition, a large body of evidence shows that
dyslexics are more likely to have a problem with their low-level sensory processing
skills, especially as it relates to visual motion. This being the case, the view that a
visual deficit could possibly play a causal role in dyslexia is not unreasonable since
fundamentally reading involves moving our eyes between words and visually
processing symbols that make up written language. Furthermore, any problem at the
oculomotor or perceptual level is likely to have a downstream effect.
This view however has not been universally accepted into mainstream thinking due to evidence showing
that dyslexia is strongly linked to a phonological deficit along with various studies that fail to support vision
as a contributing factor. Indeed it is commonly believed that the erratic eye movements observed in
dyslexics are primarily due to difficulty with language and do not have a visual basis at all. Most reviews on
this subject present evidence on both sides of the argument but usually conclude by either favoring the bias
of the reviewer or more likely, remain inconclusive except to confirm a phonological link. Those against
vision playing a role in dyslexia often raise the lack of randomized trials but it is unusual to find a review that
critically analyzes the methods of studies that fail to find a visual link.
It is the goal of this presentation to provide such a review and by doing so show using examples that studies
claiming a negative result may in fact suffer from a lack of specificity (ie. how accurate the visual test is for
ruling out dyslexia) or lack of sensitivity (ie. how accurate the visual test is for identifying dyslexia). In doing
so, it is hoped to show that vision may play a more important role in dyslexia than many experts currently
hold to be the case and that it may even play a causal role.

3. Lack of Specificity: Example 1
• The first example refers to a group of eye tracking studies that were done in the eighties showing that
saccades are not affected in dyslexia.1-
• Brown B, Haegerstrom-Portnoy G, Yingling C.D, Herron J, Galin D, Marcus M. Predictive eye movements
do not discriminate between dyslexic and control children. Neuropsychologia 1983, 21(2):121-128
• Stanley G, Smith G, Howell E.A. Eye-movements and sequential tracking in dyslexic and control children.
British Journal of Psychology 1983, 74:181-187.
• Black JL, Collins DW, De Roach JN, Zubrick S. A detailed study of sequential saccadic eye movements for
normal –and poor reading children. Perceptual Motor Skills 1984, 59(2):423-434.
• Common to all these studies is that they compare saccade latencies (reaction times) of both dyslexic and
non-dyslexic students between the ages of 9 to 13 years.
• Although it is not explicitly stated that they are testing “voluntary saccades” (the eye movements required
for reading) their methods indicate this is probably the case.
• In all of these studies no significant difference was found in saccade latencies.
• Is this sufficient however to conclude that saccades are the same in dyslexics as non-dyslexics? What
about other aspects of eye movements?
• Contrast these with the following study comparing dyslexics with non-dyslexics for saccade accuracy.
• Biscaldi M, Fischer B, Hartnegg K. Voluntary saccadic control in dyslexia. Perception 2000, 29:509-521.
• Although these investigators confirm the above studies (ie. no difference in saccade latencies compared to
age matched controls in the 9 to 13 year old age group), they find a very significant difference in saccade
accuracy (ie. the number of errors OR misses made) during an anti-saccade task which requires the
subjects to make voluntary eye movements (see slides to follow).
• This has subsequently been confirmed by other studies.1-2

4. Lack of Specificity: Example 1
• The graphs below show no difference in the reaction times or latencies of either pro-saccade (reflex) or
anti-saccade (voluntary) tasks for students 9 to 13 years of age, the age group MOST OFTEN USED for
testing saccades. Differences do occur however in the younger or older age range.
Graphs of Saccade Reaction Times for Dyslexics & Controls versus Age (from Fischer)

5. Lack of Specificity: Example 1
• By contrast, the graphs below show a very significant difference in saccade accuracy between dyslexics
and non-dyslexics across age as measured by the number or errors or the number of misses (errors that
were failed to be corrected).
Graphs of Saccade Errors & Misses for Dyslexics & Controls versus Age (from Fischer)
Misses in Antisaccade Task

6. Discussion
• The study by Biscaldi, Fischer & Hartnegg clearly shows significant differences in eye saccades between
dyslexics and non-dyslexics on an anti-saccade task.
• The question of saccades is primarily an issue for voluntary eye control rather than saccade latencies
although differences in saccade latencies can still be observed at the extreme ends of the range for age.
This finding is consistent with the notion that reading is a consciously directed task that requires a high
level of eye control.
• In addition to this, other oculomotor deficits have also been reported to occur in
students with dyslexia and general learning difficulties with pursuit eye movements,
binocular stability and convergence.3-12
• Although it could be argued that the anti-saccade task is a test of attention rather
than eye movements, a case could be made that it is a test of both. That is to say,
the type of attention required to make an anti-saccade is highly specific and thus
could be considered part of the eye movement itself. Evidence for this comes
from Fischer who found that training reflex saccades on a hand-held device
(like the one shown in the picture) does not affect voluntary saccades and
that training voluntary saccades using the same device does not affect reflex
saccades, however in both cases the training significantly changes the eye movement being trained.13
• Regarding pursuit eye movements, these are often ignored despite studies showing differences exist
between dyslexics or poor readers and proficient readers in younger children. This is probably because
pursuits are not thought of as being important for reading, however it should be noted that pursuits
and saccades share a lot of the same neural architecture and possibly even the same neurons so it
should not be too surprising to see that saccade errors also occur in dyslexia!14
• See references on the next slide.

7. REFERENCES
1. Fukushima J, Tanaka S, Williams JD, Fukushima K. Voluntary control of saccadic and smooth-pursuit eye movements in children with learning
disorders. Brain and Development 2005, 27(8): 579-588.
2. Lukasova K, Silva I, Macedo E. Impaired oculomotor behaviour of children with developmental dyslexia in antisaccades and predictive
saccades tasks. Frontiers in Psychology 2016, 7:987
3. Callu D, Giannopulu I, Escolano S, Cusin F, Jacquier-Roux M, Dellatolas G. Brain and Cognition 2005,58(2):217-225
4. Eden G, Stein J, Wood H. Differences in eye movements and reading problems in dyslexic and normal children. Vision Res 1994,34(10): 1345-
1358
5. Fischer B, Hartnegg K. Stability of gaze control in dyslexia. Strabismus 2000:8:119-122
6. Stein JF, Fowler MS. Unstable binocular control in children with specific reading retardation. J Res Reading 1993, 16:30-45.
7. Kapoula Z, Ganem R, Poncet S, Gintautas D, Bremond-Gignac D. Poor binocular yoking of the saccades independently from reading in dyslexic
children. Perception 2006: 35 ECVP Abstract Supplement
8. Bucci MP, Bremond-Gignac D, Kapoula Z. Poor binocular co-ordination of saccades in dyslexic children. Graefe’s Archive for Clinical and
Experimental Ophthalmology 2007, On-Line First.
9. Raghuram A, Gowrisankaran S, Swanson E, Zurakowski D, Hunter D, Waber D. Frequency of Visual Deficits in Children With Developmental
Dyslexia JAMA Ophthalmol 2018,136(10):1089-1095
10. Dusek W, Pierscionek BK, McClelland JF. A survey of visual function in an Austrian population of school-age children with reading and writing
difficulties. BMC Ophthalmology 2010, 10(16).
11. Shin HS, Park SC, Park CM. Relationship between accommodative and vergence dysfunctions and academic achievement for primary school
children. Ophthalmic & Physiological Optics 2009, 29(6):615-624.
12. Grisham D, Powers M, Riles P. Visual skills of poor readers in high school. Optometry 2007, 78(10):542-549.
13. Fischer B, Hartnegg K. Effects of visual training on saccade control in dyslexia. Perception 2000, 29(5):531-42.
14. Krauzlis RJ. The control of voluntary eye movements: new perspectives. Neuroscientist 2005, 11(2):124-37.

8. Lack of Specificity: Example 2
• The next example is a paper which tests a range of ocular skills in dyslexics:
• Creavin AL, Lingam R, Steer C, Williams C. Ophthalmic abnormalities and reading impairment. Pediatrics.
2015;135:1057–65
• This was a large prospective population study in the UK, on students 7 to 9 years of age (n=5822) of whom
651 were dyslexic (11%).
• The visual skills tested were strabismus (squint), motor fusion, sensory fusion at distance, refractive error,
amblyopia, convergence, accommodation and contrast sensitivity.
• Results: The study found no association between dyslexia and the visual skills described and concludes
that this provides further evidence that dyslexia is a linguistic and not a visual disorder.
• By comparison consider the following study which also assesses ocular skills in dyslexics:
• Raghuram A, Gowrisankaran S, Swanson E, Zurakowski D, Hunter D, Waber D. Frequency of Visual
Deficits in Children With Developmental Dyslexia. JAMA Ophthalmol. 2018;136(10):1089-1095
• This is a much smaller study which compares oculomotor skills in 29 children with Developmental Dyslexia
(average age 10.3 years) and 33 typically developing (TD) children
• Visual skills tested were : vergences (amplitude, fusional ranges, and facility), accommodation (amplitude,
facility, and accuracy), and ocular motor tracking (Developmental Eye Movement test and Visagraph eye
tracker).
• Results: It found that 23 children in the DD group (79%) and 11 children in the TD group (33%) had deficits
in 1 or more domain of visual function (P < .001). This would suggest that visual deficits are more likely to
be present in dyslexia.

9. Discussion
• Although there is a big difference in the number of students tested in these studies, they are typical of
studies in the literature showing either no difference or a very large difference. These studies were
selected as they are both recent studies published in reputable medical journals.
• According to Fischer who has assessed around 3000 German dyslexic students (the highest number found
by this author), 99.9% of these have either a visual OR auditory deficit and between 80-92% have a
problem with BOTH depending on age.1
• How can there be such large discrepancies? The answer most likely is found in the type of visual tests
administered. That is to say; there are certain types of visual skills which are more likely to be deficient in
dyslexics than others.
• In order to have an intelligent conversation regarding the role of vision in dyslexia therefore, it is important
to identify what type of visual skills one is referring to.
• The types of visual skills most often observed to be a problem in dyslexia are those requiring rapid visual
processing. These are often linked to visual motion and visual spatial attention and are associated with the
Magnocellular Visual Pathway, a distinct anatomical pathway in the brain that originates in the eye and
extends from the primary visual cortex to a number of cortical regions in the brain critical for reading via
the so called “dorsal route”.
• In the case of reading this likely supports skills such as saccadic eye control, binocular stability, visual
counting, the visual span and spatial attention all of which entail rapid visual processing prior to word
identification. These skills may be considered as indices of magnocellular function.
REFERENCES
1. Stein J, Kapoula Z. Visual Aspects of Dyslexia, 2012. Oxford University Press, 2012. Chapter 2: Subitizing, Dynamic Vision, Saccade and Fixation
Control in Dyslexia.

10. Lack of Specificity: Example 3
• The next example will look from the perspective of training visual skills in dyslexia:
• Peyre H, Gerard CL, Dupong Vaderhorst I, Lemoussu C, Bui Quoc E, Deloeme R, Bucci MP. Computerized
oculomotor training in dyslexia: A randomized, crossover clinical trial in a pediatric population.
Encephale, 2018 Jun;44(3):247-255.
• This study is a crossover randomized trial involving 11 dyslexic children ages 7 to 12 years
• It involved computer-based oculomotor training in the following areas: voluntary AND reflex saccades,
vergence training and visual attention. It measured the outcomes at 3 & 6 months.
• Results: The results showed no effect of oculomotor training on reading skills but a positive effect on
writing and several other cognitive skills.
• Compare this with a study that involved targeted saccade training:
• Fischer B, Hartnegg K. Saccade control in dyslexia: development, deficits, training and transfer to
reading. Optom & Vis Devel 2008, 39(4):181-190
• Training involved doing a customized saccadic training programme
over a period of weeks for 8 to 13min/day using a hand-held device
(as shown previously).
• Fixation & reflex training were ONLY given if indicated from eye
tracking analysis but mostly involved anti-saccade training.
• Both the dyslexic treatment and control group also received
6 weeks of reading instruction.
• Results: This resulted in about a 50% drop in reading errors by the
treatment group compared to no more than 20% by the control group
(see graph to right, supplied by Fischer).

11. Discussion
• In the study by Peyre et al there was a positive effect of training on writing skills but there was no effect on
reading outcomes. By comparison, Fischer showed a significant reduction in reading errors of around 50%
but did not consider writing skills.
• One explanation for this difference is the targeted approach used by Fischer, especially as it relates to
training reflex saccades since if used indiscriminately this could have a negative effect on reading. If one
considers that for a normally sighted person fixation is an active process that over-rides reflexive eye
movements then when a reflex saccade is made this requires turning off fixation to enable the reflex to
occur. If done repeatedly this could strengthen the reflex saccade (shown by shorter saccade latencies)
and as a consequence weaken fixation.
• Weaker fixation means less visual information is being processed per look during the fixation phase of
reading thereby leading to a possible drop in reading performance.
• This mechanism is described by Fischer in the Optomotor Cycle (see next slide).1
RELEVANCE TO COMPUTER GAMES
• A similar problem may also occur when playing action computer games, especially for younger students
who have rapidly developing oculomotor systems. There are presently a number of studies to show that
playing action computer games improves visual processing speed in dyslexics, but nobody appears to have
considered whether this has any negative effects.
• As a side note, the author conducted a small pilot study (unpublished) which looked at the effect of arcade
style computer games in primary aged children with learning difficulties over a three month period. It
found that some visual skills such as processing speed significantly improved but that reading
comprehension did not (in fact it went backwards) compared to those who did not play arcade style
computer games.
• It is therefore essential to understand the nature of the visual intervention given and to consider
whether or not there may be any negative effects.

12. The Optomotor Cycle
• In the Optomotor Cycle proposed by Fischer, fixation normally overrides reflex eye movements but when a
reflex eye movement is required to be made then fixation must be switched off. The ability to fixate or not
is under frontal lobe control whilst the reflex saccade is a mid-brain function that involves the superior
colliculus.
• Training reflexive eye movements weakens fixation (by reducing the reflex reaction time). Training fixation
however strengthens fixation (by increasing the reflex reaction time). This is critical as there is an optimal
fixation period that occurs when reading (about 300ms) to process information for comprehension.
• By comparison, the anti-saccade task relates directly to reading eye movements. Training anti-saccades
(voluntary saccades) does not affect the reaction times of reflex saccades in the optomotor cycle and
therefore is unlikely to negatively impact upon fixation.
REFERENCES
1. Visual Aspects of Dyslexia. Subitizing, Dynamic Vision, Saccade and Fixation Control in Dyslexia. Oxford University Press, 2012.
The Optomotor Cycle (supplied by Fischer)

13. Lack of Sensitivity: Example 1
• The next examples will shift to considering the sensitivity of eye tracking studies in dyslexia:
• Hutzler F, Kronbichler M, Jacobs A, Wimmer H. Perhaps correlational but not causal: No effect of dyslexic
readers’ magnocellular system on their eye movements during reading. Neuropsychologia 2006, 44:637-
648
• This study compared dyslexics students with non-dyslexic students in a visual search task involving
consonant strings (CS) and pseudoword strings (PW). The average age of the students was around 13 yrs.
• Results:
A. Consonant String task: Dyslexics = Non-dyslexics
B. Pseudoword task: Dyslexics worse than Non-dyslexics
• Conclusion: “The absence of any difference between in eye movement patterns during the string
processing furthermore suggests that the divergent eye movement patterns of dyslexic readers during the
pseudoword reading solely reflect their difficulties in the reading process”
• Prado C, Dubois M, Valdois S. The eye movements of dyslexic children during reading and visual search:
impact of the visual attention span. Vision Res 2007 , 47(19):2521-30
• This study compares dyslexic students with non-dyslexic students in a reading task and a visual search task
using age appropriate text. The average age of the students was 11 years.
• Results:
A. Reading task (paragraph of text): Dyslexics worse than Non-dyslexics
B. Visual Search task (paragraph of text where vowels were replaced with consonants and asked to count
the number of ‘r’s occurring in the text: Dyslexics = Non-Dyslexics
• C. Dyslexics had a reduced visual span = a possible reason for their atypical eye movements
• Conclusion: “Atypical eye movements do not seem to result from a primary oculomotor disorder” but a
reduced visual span could be a factor.

14. Discussion
• It has long been observed that dyslexics make atypical eye movements when reading. The question is
whether this represents a problem with oculomotor control or language processing since both are required.
In the studies cited, erratic eye movements were observed on the visual search tasks that involved reading
text or pseudo words but NOT once language was removed from the task.
• A problem with both these studies is they fail to consider whether adding a linguistic component creates a
higher cognitive load on saccade control and hence the breakdown in eye movements observed during
reading may reflect the higher cognitive load placed on the oculomotor system rather than difficulty with
language itself. This idea has been investigated by Larter et al (2004) who found that spatial loading
predicted reading performance.1 The question then is, if the oculomotor test was made more difficult
independently of language would dyslexics still perform equally well? The earlier study cited in this
presentation by Fischer that uses an anti-saccade task would suggest not. At the clinical level this idea can
easily be observed directly by asking a child with learning difficulties to vocalize whilst performing an eye
movement task. The cognitive load becomes too high and the skill starts to breakdown.
• An alternative to increasing the cognitive load would be to be perform the SAME test on students in a
younger age group (now that we have higher resolution head free eye trackers available) since it can be
shown that the visual skills in question undergo a steep development during the primary school years and
are therefore less likely to cope with the same cognitive loading.
• In both scenarios, one would expect dyslexics to perform worse. Consequently, the visual search tasks
described by these authors may simply be depicting a ceiling effect (ie. the task is far too easy for age).
• On a practical note, this could explain why a visual scanning test used by optometrists called the DEM
(Developmental Eye Movement test) does not correlate well with anti-saccade testing using an eye tracker.2
The DEM is better suited for younger students (ages 6 to 8 years) but starts to lose sensitivity from around
age 9, after which it becomes a test of rapid naming. To claim that it is NOT a test of eye tracking at all fails
to take into account the rapid development of saccades that occurs in younger children.2

15. REFERENCES
1. Larter S, Herse P, Naduvilath T, Dain S. Spatial load factor in predicting reading performance. Opthal Physiol Optics 2004, 24(5):440-49.
2. Ayton L, Abel L, Fricke T, McBrien N. Developmental eye movement test: what is it really measuring? Optom Vis Sci 2009, 86(6):722-30

16. Lack of Sensitivity: Example 2
• Having considered eye movements, could a similar problem be observed for visual processing?
• Historically there has been much debate regarding the contribution of visual information processing
deficits and linguistic processing. The work of Vellutino et al detailed below was arguably the most
influential in discrediting visual processing.
• Series 1: Memory for Familiar Symbols
Study 1 – Vellutino, Steger, Kandel 1972
Study 2 – Vellutino, Smith, Steger, Kaman 1975
• Series 2: Memory for Unfamiliar Symbols
Study 3 – Vellutino, Pruzek, Steger, Meshoulam 1973
Study 4 – Vellutino, Steger, Kaman, DeSetto 1975
Study 5 – Vellutino, Steger, DeSetto, Phillips 1975
• This included a series of studies comparing poor readers (2 or more years behind in reading) with good
readers (at or above age level) on a number of tasks involving verbal and non-verbal stimuli.
• All students were taken from a sample pool with a Verbal or Performance IQ of 90 or above.
• In Series 1 students were presented with real & scrambled words (3 to 5 letters), numbers (3 to 5 digits)
and geometrical designs.
• For Study 1 the average age was 11.9 years (from 9 to 15 comprising the older group) and in Study 2 it
was between 7 to 8 years of age (ranging from 6 to 8) comprising the younger group.
• The visual component of the testing in both studies required students to copy (recall) the stimuli from
memory.
• Results: In Study 1, poor and normal readers averaged 96% correct for geometrical designs and on the
number strings all children were 100% accurate. On the 3 letter strings normal readers were 91% accurate
and poor readers were 62% accurate but this was not deemed significant. On the 4 to 5 letter strings there
was a “floor effect” with results being too low in both groups so as to be meaningful.

17. • In Series 2 students were presented with Hebrew Symbols (unfamiliar stimuli) 3 to 5 characters in length
which were displayed for between 3 to 5 seconds and then asked to recognize the symbols.
• The age of the students in Study 3 was 10-13 years (the older students). In Study 4 the students averaged
over 8 years (the high end of the young age) and in Study 5 the average age was 7.9 years (the younger
students).
• Results: In all three studies it was found that poor readers were not statistically different to normal
readers despite students that were poor readers performing worse. In the youngest group the means were
29% correct for the poor readers versus 48% correct for the good readers. The sample size of each group
ranged from 11 to 14 students depending on the study.
• By comparison, consider tasks where visual perceptual testing is more targeted:
• Bosse ML, Tainturier MJ, Valdois S. Cognition. 2007, Aug;104(2):198-230.
• The visual attention span was defined as the number of distinct visual elements that can be processed at
once. A five letter consonant string (eg. R H S D M) is flashed for 200ms at the center of a screen and
participants must report either the whole letter string or a single cued letter (partial report). This has been
shown to measure parallel processing (or the visual attentional window) rather than memory load. In the
study above, samples of French and British dyslexic children (n=68 & n=29) were assessed with a mean
age of 9 to 11 years and compared them with age matched controls after controlling for age, IQ etc.
• Results: Both VA span and phonological processing skills were found to independently contribute towards
predicting reading performance. The number of dyslexic children with a VA span disorder was at least as
high as the number with a phonological disorder.
• The authors conclude that their study supports a multi-factorial view of dyslexia and that in many cases
the reading disorder does not appear to be phonological.

18. • Fischer B, Gebhardt C, Hartnegg K. Optom & Vis Dev 2008:39(1):24-29
• Subitizing is the ability to recognize a number of briefly presented items (small circles) in 100ms without
actually counting them. As the number of items is increased it requires counting the items which relies on
visual memory of the test pattern. This capacity can be measured in terms of both accuracy and speed of
response.
• The above study compared 219 control subjects with 156 subjects that had difficulty with arithmetic for an
age range of 7 to 17 years.
• Results: A significant difference between students with dyscalculia and no dyscalculia was found for
between 40% to 78% of subjects (increasing with age).
• Preliminary data also showed that up to 60% of dyslexic children also suffer from deficits in subitizing and
counting. The link to dyslexia has later been confirmed by on-going testing.
• The graphs below show the effect of subitizing /visual counting on the average reaction time per item (2
to 8 items) and accuracy of responses.
Subitizing in Dyslexics vs Controls. (Graphs supplied by Fischer)

19. Discussion
• The studies by Vellutino have been criticized by Willows et al due to a number of deficiencies including
procedural weaknesses, statistical problems due to ceiling/floor effects and conceptual problems.1-2
• They argue that the first series of studies suffer from ceiling effects (the tests are too easy) except on the 5
letter word task where a statistically significant difference was found between poor readers (53%) and
normal readers (84%) but this was explained away.
• The second series of studies suffered from floor effects (the tests are too difficult) resulting in low correct
response rates in which 15% of the accuracy could be explained by chance alone.
• Another criticism is the need for tighter parameters on the presentation times (which had a duration of 3
to 5 seconds) and also a limit on the response time (or else measuring the response time at all) since this
may have revealed a difference where no difference in accuracy was found.
• In contrast, the studies by Boss et al (2007) and Fischer et al (2008) which measure the visual span and
subitizing respectively have more defined parameters. In both cases they show a clear difference in visual
processing between dyslexics and non-dyslexics.
REFERENCES
1. Willows D, Kershner J, Corcos. Visual Processing and Visual Memory in Reading and Writing Disabilities: A Rational for Reopening a “Closed
Case”. Paper at American Educational Research Association, April 1986.
2. Willows D, Kruk R, Corcos E. Visual Processes in Reading and Reading Disabilities. Digital Printing 2009.

20. Lack of Sensitivity: Example 3
• The most recognized tests of magnocellular function involve measuring coherent motion and temporal
contrast sensitivity:
• Ramus F, Rosen S, Dakin S, Day B, Castellote, White S, Frith U. Theories of developmental dyslexia:
insights from a multiple case study of dyslexic adults. Brain 2003, 126(4):841-65
• This was a comprehensive study that involved testing 16 dyslexic and 16 control students of university age.
It involved a full battery of psychometric, phonological, auditory, visual and cerebellar tests.
• The visual tests targeted magnocellular function and included testing contrast sensitivity (with low spatial
frequency and counter phased flicker at low luminance), speed discrimination and coherent motion (the
level of coherence required to observe that dots are moving in a particular direction on a screen).
• Results:
• No difference was found in contrast sensitivity, speed discrimination or coherent motion between
dyslexics and non-dyslexics.
• In direct contrast there was a strong link (100%) with phonological skills.
• The authors conclude this provides strong support for a phonological basis to dyslexia and not a visual
basis.

21. Lack of Sensitivity: Example 3
• Compare this with a study using visual motion testing in younger students:
• Gori S, Seitz A, Ronconi L, Franceschini S, Facoetti A. Multiple causal links between magnocellular-
dorsal pathway deficit and developmental dyslexia. Cereb Cortex 2016, 26(11):4356-4369.
• This involved multiple experiments which included coherent motion in 15 dyslexic students with an
average age of 11 years matched to controls of similar IQ and age plus a younger control group (age
8.46 years) functioning at a similar reading level.
• In addition they performed a longitudinal study over 3 years that involved testing coherent motion on
72 children starting at the pre-school level (ie. pre-readers).
• Results:
A. Their findings showed that coherent motion testing significantly discriminated between dyslexic
and typically developing readers (including the younger non-dyslexic students) at all levels of
coherence.
B. The longitudinal study showed that coherent motion testing successfully predicted poor readers
(n=12) compared to typical readers (n=60)
• The authors conclude that their study (which also included a treatment protocol that involved
training coherent motion) demonstrates a causal role of magnocellular pathways in developmental
dyslexia.
• Side Note: The longitudinal study also included a visual spatial search task which significantly
predicted future reading acquisition. This is consistent with the view posited earlier that a non-
linguistic visual search task is more likely to a show a significant difference in younger children
compared to older children due to the cognitive load being relatively higher.

22. Discussion
• Although many studies using coherent motion and low contrast moving sinusoidal gratings report a
difference in sensitivity between dyslexics and non-dyslexics1 others such as Ramus et al have not.2-4
• One possible problem with the Ramus experiment was the dot density used was quite high, at about 18.75
dots/degree. This has the effect of increasing the total number of motion signals to the brain which
according to Conlon (2013) may reduce the sensitivity of the test.5 This may also explain the lack of
significant findings by Hill & Raymond (2002) who used 45 dots/degree.3 By comparison Gori et al used
just 10 dots/degree with Conlon claiming that an ideal number is less than 9 dots/deg.
• Conlon also argues that the coherence test may be made more difficult by decreasing the contrast of the
signal dots relative to the background dots. This causes the higher contrast background dots to be
preferentially processed thereby increasing the cognitive load of the test.
• In the case of the Ramus study the contrast of both the background and stimulus/signal dots was set quite
low (white on a grey background) which made it more difficult to see the dots compared with other
studies using higher contrast dots (closer to 100%) for coherent motion testing.
• Another aspect to consider is the way in which the test is administered. For example, in the Gori study the
coherence threshold was fixed (at say 30 to 40%) and the percentage of errors was measured over a fixed
number of trials (eg. 80 trials). This makes it more analogous to the anti-saccade and subitizing tests of
Fischer described earlier and may provide a better measure as threshold values are likely to change with
age due to the long development of the magnocellular pathways.
• Regarding temporal contrast sensitivity testing, this is much more likely to be effective for poor readers
when done against a stationary textured background (especially with multiple spatial frequencies)
compared to no background which can reduce the sensitivity of the test.6

23. • Having a background that uses multiple spatial frequencies effectively increases the cognitive load of the
test (due to the higher figure ground discrimination required) thereby exposing any underlying weakness
in the magnocellular system. Doing this with a moving background could reduce the sensitivity of the test
in much the same way as using excessive moving dots may for the coherence motion test.
• Regarding training, multiple studies have shown that training random coherent motion discrimination can
improve reading outcomes confirming the link between magnocellular processing and dyslexia.7-10 In
addition, Lawton (2016) has demonstrated the efficacy of using moving sinusoidal gratings where the
direction must be detected against a stationary sinusoidal background thus creating a challenging figure
ground task.6
• An alternative approach may be to combine training of critical cortical skills such as saccades or the visual
span (top-down) with a magnocellular component such as using a visual motion background (bottom-up).
REFERENCES
1. Stein J, Kapoula Z. Visual Aspects of Dyslexia, 2012. Oxford University Press, 2012.
2. Williams M, Stuart G, Castles A, McAnally K. Contrast sensitivity in subgroups of developmental dyslexia. Vision Res 2003 43(4):467-77
3. Hill G, Raymond J. Deficits of motion transparency perception in adult developmental dyslexics with normal unidirectional motion senstivity.
Vision Res 2002, 42(9):1195-203
4. Amitay s, Ben-Yehudah G, Banai K, Ahissar M. Disabled readers suffer from visual and auditory impairments but not from a specific magnocellular
deficit. Brain 2002 125(10):2272-85.
5. Conlon E, Lilleskaret G, Wright C, Stuksrud A. Why do adults with dyslexia have poor global motion sensitivity? Front Hum Neurosci 2013,12:859.
6. Terri Lawton. Improving dorsal stream function in dyslexics by training figure/ground motion discrimination improves attention, reading fluency,
and working memory. Front Hum Neurosci 2016; 10: 397.
7. Solan H, Shelley-Tremblay J, Hansen P, Silverman M, Larson S, Ficarra A. M-cell deficit and reading disability: a preliminary study of the effects of
temporal vision-processing therapy. Optometry 2004, 75(10):640-650.
8. Gori S, Seitz AR, Ronconi L, Franceschini S, Facoetti A. Multiple causal links between magnocellular-dorsal pathway deficit and
developmental dyslexia. Cerebral Cortex 2015, Sep 22;1-14.
9. Qian Y, Bi H. The effect of magnocellular-based visual-motor intervention on Chinese children with developmental dyslexia. Front. Psychol. 2015,
doi.org/10.3389/fpsyg.2015.01529
10. Ebrahimi L, Pouretemad H, Khatibi A, Stein J. Magnocellular based visual motion training improves reading in Persian. Sci Rep 2019, 9:1142

24. Other Considerations
• Apart from specificity and sensitivity, another aspect to consider is the interpretation of the study findings.
An example is the study by Olulade et al (2013) which found that dyslexic students with a mean age of
10.4 years exhibited less visual brain activity (in area V5/MT) during visual motion processing using fMRI
compared to age matched controls but not when matched to younger students (mean age of 7.5 years)
with similar reading ability.1 Further, they found that phonological-based reading intervention increased
V5/MT activity demonstrating that visual activity is mobilized through reading. Together this was provided
as evidence that visual motion deficits are caused by a lack of reading practice.
• Although this is one way of interpreting the evidence, another possibility to consider is that a deficit in the
visual system itself explains the reduced visual activity. Such a view may be difficult to accept given the
widely held belief that dyslexia is primarily a language disorder. The fact that visual activity was similar to
that of younger students the same reading age is a correlational finding and does not indicate whether
this can be attributed to language or vision or indeed whether there may be some other factor affecting
the development of both. The finding that reading intervention improves visual activity is not too
surprising since there may be a degree of reciprocal effect in much the same way as with phonological
processing and reading. In fact, Solan has shown in a previous crossover design study that comprehension
therapy can significantly improve visual attention and vice versa.2
• That said, the idea of a language deficit causing reduced activity in V5/MT would imply an upstream effect
of language on vision whereas the alternative is far more likely. This can be avoided however by viewing
the findings in terms of a lack of reading practice as per the Olulade study. There is evidence to suggest
however that lack of reading practice is not the reason. A study by Flint & Pammer (2019) comparing
illiterate non-dyslexic adults (never learned to read) with dyslexics of similar reading ability found that
illiterate adults have significantly better visual motion skills (using coherent motion) leading the authors to
conclude that reduced visual skills in dyslexics are not caused by lack of reading practice.3

25. • This raises the question that if dyslexia was caused by a visual deficit or at least in part, could this be due
to a delay in the development of the magnocellular pathway? If so, how would this occur? One possible
explanation comes from the work of French physicists’ Floch & Ropar (2017) who claim a difference in the
pattern of retinal receptors between dyslexics and non-dyslexics leading to a lack of sensory eye
dominance.4 The binocular rivalry resulting from a lack of sensory eye dominance could potentially impair
the development of the magnocellular pathways critical for rapid visual processing which enables early
word recognition (shape and size of a word) as well as where to position the eyes and to hold a steady
gaze during fixation subsequent to language processing.
• This is not difficult to envisage given the long time that it takes for magnocellular linked skills to reach
their peak level of development (mid to late teens) making them susceptible to delays whether they have
a genetic or non-genetic origin. In either case, similar treatments may apply that could involve stimulating
the pathway directly (such as monocular occlusion, intermittent central suppression therapy, using visual
motion with selected spatial frequencies or moving dots and tinted lenses) or indirectly by targeting the
cortical processes critical for rapid visual processing (such as optometric vision therapy and or computer-
based therapies). The ideal solution may involve a combination of both. This is the subject of another
review, but evidence to date is promising.
REFERENCES
1. Olulade O, Napoliello E, Eden G. Abnormal visual motion processing is not a cause of dyslexia. Neuron 2013, 79(1):180-90
2. Solan H, Larsen S, Shelly-Tremblay J. Role of visual attention in cognitive control of oculomotor readiness in students with reading disabilities. J Learn
Disabil 2001, 34(2):107-18.
3. Flint S, Pammer. It is the egg, not the chicken; dorsal visual deficits present in dyslexia are not present in illiterate adults. Dyslexia 2019, 25(1):69-83
4. Le Floch A, Ropars G. Left-right asymmetry of the Maxwell spot centroids in adults with dyslexia. Proc Biol Sci 2017, 25;284.

26. Conclusion
• To resolve the debate on whether vision is a contributing factor in dyslexia one needs to critically assess
studies showing that it does not play a role. Unfortunately, many reviews only compare studies for and
against the role of vision with little or no critical analysis.
• When examined carefully however it can be shown that studies finding no visual link may suffer from poor
specificity or sensitivity. In addition, the strongly held view that dyslexia is a language problem may be
influencing the interpretation of results that might also suggest a vision problem.
• Poor specificity is relatively easy to identify since visual skills associated with magnocellular processing are
likely to involve rapid temporal or spatial processing and be located on the dorsal route in the brain such
as the right parietal cortex for spatial processing or the frontal eye fields in the case of saccade control.
• Poor sensitivity can arise due to a number of reasons but especially when the visual task does not have an
equivalent non-linguistic cognitive load (ie. the task is too easy) compared to the more difficult reading
task thus leading to ceiling effects.
• Alternatively, if the same task showing no significant difference is given to a younger age group this is likely
to show as significantly different because the magnocellular skills in question are less developed.
• Finally, given that 1) Phonological skills accounts for less than 50% of the variance in reading ability and
that the number of dyslexic students with a visual processing disorder (such as a reduced visual span) is at
least as high as the number with a phonological disorder1-4, 2) Dyslexics have reduced visual activity as
shown by functional MRI along with various sub-optimal skills linked to the visual magnocellular
pathway5,6, 3) Tests failing to show a difference between dyslexics and non-dyslexics may be flawed due to
lack of specificity & sensitivity and 4) The growing number of studies that show training of magnocellular
skills improves learning outcomes7-15, then one must conclude that a visual link is reasonably supported in
the scientific literature and that this should be considered as part of the multi-disciplinary
approach to managing dyslexia.

27. • Further, given the long development of the magnocellular pathways this could be influenced by both
genetic and non-genetic factors. As such, whether the learning delays are specifically associated with
dyslexia or considered to be more general in nature then similar treatments may apply which involve
either stimulating the magnocellular pathway directly (bottom-up approach) or targeting its ancillary
cortical processes (top-down approach) such as optometric vision therapy and or computer-based
therapies.
• Finally, this review highlights the need for large scale studies that measure how visual skills change with
age such as the studies published by Fischer. If greater emphasis had been given to measuring the visual
development of dyslexics and non-dyslexics earlier then it is possible many of the problems described in
this review may have been avoided.
REFERENCES
1. Mann VA, IY Liberman. Phonological awareness and verbal short-term memory. J Learn Disabil. 1984, 17(10):592-9
2. Wagner R. K. Changing relations between phonological processing abilities and word-level reading as children develop from beginning to skilled readers:
a 5-year longitudinal study, 1997. Dev. Psychol. 33, 468–479.
3. Bosse ML, Tainturier MJ, Valdois S. Developmental dyslexia: the Visual Attention Span hypothesis. Cognition 2007, 104:198-230
4. Germano G, Reilhac C, Capellini SA, Valdois S. The phonological and visual basis of developmental dyslexia in Brazilian Portuguese reading children. Front
Psychol 2014, 5 (1169):1-11.
5. Olulade O, Napoliello E, Eden G. Abnormal visual motion processing is not a cause of dyslexia. Neuron 2013, 79(1):180-90
6. Shaywitz S. E., Shaywitz B. A., Pugh K. R., Fullbright R. K., Constable R. T., Mencl W. E., et al, 1998. Functional disruption in the organization of the brain
for reading in dyslexia. Proc. Natl. Acad. Sci. U S A 95, 2636–2641.
7. Solan H, Larsen S, Shelly-Tremblay J. Role of visual attention in cognitive control of oculomotor readiness in students with reading disabilities. J Learn
Disabil 2001, 34(2):107-18.
8. Ebrahimi L, Pouretemad H, Khatibi A, Stein J. Magnocellular based visual motion training improves reading in Persian. Scientific Reports, 2019 (9):1142.
9. Jafarlou F, Jarollahi F, Ahadi M, Sadeghi-Firoozabadi V, Haghani H. Oculomotor rehabilitation in children with dyslexia. Med J Islam Repub Iran 2017,
24(31):125.
10. Leong D, Master CL, Messner LV, Pang Y, Smith C, Starling AJ. The effect of saccadic training on early reading fluency. Clinical Pediatrics 2014, 53(9):858
11. Dodick D, Starling AJ, Wethe J, Pang Y, Messner L, Smith C, Master CL, Halker-Singh RB, Vargas BB, Bogle JM, Mandrekar J, Talaber A, Leong D. J. The
effect of in-school saccadic training on reading fluency and comprehension in first and second grade students. Child Neurol 2017,
32(1):104-111.

28. REFERENCES
12. Ebrahimi L, Pouretemad H, Khatibi A, Stein J. Magnocellular based visual motion training improves reading in Persian. Sci Rep 2019, 9(1): 1142.
doi:10.1038/s41598-018-37753-7.
13. Gori S, Seitz AR, Ronconi L, Franceschini S, Facoetti A. Multiple causal links between magnocellular-dorsal pathway deficit and developmental
dyslexia. Cerebral Cortex 2015, Sep 22;1-14.
14. Lorusso ML, Facoetti A, Cattaneo C, Pesenti S, Galli R, Molteni M, Geiger G. Training visual-spatial attention in developmental dyslexia. In:
Dyslexia in Children: New Research 2006, pp 143-160.
15. Terri Lawton. Improving dorsal stream function in dyslexics by training figure/ground motion discrimination improves attention, reading fluency,
and working memory. Front Hum Neurosci 2016; 10: 397.

29. Acknowledgment
This review acknowledges the work of Professor Burkhart Fischer and his colleagues
at the Optomotor Laboratory at the Freiburg University who spent many years
collecting a large database showing how oculomotor and
sensory processing skills change with age.
Stuart Warren & Prof Burkhart Fischer